JPH0468763A - Cutout mask generating method - Google Patents

Abstract

PURPOSE:To obtain a synthesis picture with quality of a contour line close to that of an original picture by substituting a picture element in an area for a desired picture cutout to an evaluation function representing the possibility of an optional color belonging to a background and an evaluation function representing the possibility of an optional color belonging to a real body and deciding the synthesis ratio. CONSTITUTION:An evaluation function is generated to each training area based on a picture element of each training area to be inputted. Then the evaluation function is used to calculate the density of a mask as to each picture element in a cutting area thereby generating a mask data having gradation. A color of each picture element (picture element value) in the cutting area is substituted to each evaluation function, the evaluation function whose value is a minimum among the evaluation functions belonging to the background is used as a distance X from the background and similarly, the evaluation function whose value is a minimum among the evaluation functions belonging to the real body is used as a distance Y from the real body. The distances X, Y are used to calculate a parameter theta of the mask density of the relevant picture element.

Description

[Detailed description of the invention] [Industrial application field]

The present invention relates to a method for creating a cutout mask for creating a composite image by performing arithmetic processing on original image data and performing electronic cutout and composition processing.

[Prior Art 1] In order to obtain various visual effects on a printed image, a desired background portion may be synthesized with an arbitrary substance portion, and a film of a printing plate master plate of the synthesized image may be created. When creating such an original, a binary mask cutting and combining method is generally used to create a composite image as follows. In this synthesis method, first, for example, a printing original is scanned and its pixel data is read in, and the read pixel data is binarized using an appropriate method to create binary values of on and off corresponding to each six pixels. Create mask data. Next, while referring to the binary mask data, if the data is off, the pixel data of the image that should be the real part is extracted, and if it is on, the pixel data of the image that should be the background part is extracted, thereby creating the image. Synthesize. Various methods have been known for creating mask data for use in such a binary mask cutting and combining method. In addition, there is a method for creating a gradation mask that creates a halftone area of a constant width based on binary mask data (for example, Japanese Patent Publication No. 1-31232). Based on the mask data, a band-shaped area of a constant width is created inside, outside, or on both sides of the contour line of this binary mask data, and for each pixel within the band-shaped area, the synthesis ratio of the foreground part image and the real part image is determined. Create mask data with gradation to determine. Problems to be Solved by the Invention However, the following problems have occurred in the conventional binary mask cutting and combining method. The first problem is that the boundaries of the composite image appear conspicuous, and when they appear that way, the composite image gives an undesirable impression of artificiality. For example, in a so-called natural image obtained by scanning a photographic original with a scanner, there is an area near the outline of the object that contains ambiguous pixels with a color intermediate between the color of the background and the color of the substance. Thanks to the area, the background part and the real part appear to blend together. However, since this ambiguous area does not exist near the boundary line of the image synthesized by the binary mask, the boundary line looks unnaturally distinct, giving an artificial impression. The second problem is that when there is a contour line that is diagonal to the scanning direction when reading both sides, step-like wobbling occurs in the contour line in pixel units. For example, in the natural image, the aforementioned ambiguous area exists, and the density gradually changes along the contour line near the diagonal contour line, so there is a density jump (sharp density change) between adjacent pixels. ) is small at all locations, and therefore, no step-like wobbling is visible even on diagonal contour lines. However, in the image synthesized using the binary mask, a large density jump occurs near the boundary line between pixels, so that the boundary line may appear to have a step-like jitter. It is thought that the first problem can be solved to some extent by using the gradation mask creation method, since the density gradually changes near the boundary line. However, in the gradation mask creation method, as described in, for example, Japanese Patent Publication No. 1-31232, since synthesis is performed using a density conversion rate curve stored in advance in memory, changes in gradation near the boundary line The natural image cannot be made smooth, and the density jump between pixels is still large, so the second problem still remains. On the other hand, if an attempt is made to reduce the density jump by widening the blur width, another problem arises in that the border lines appear blurred and blurred, resulting in a blurred composite image with unclear outlines. Therefore, there is a need for a technology that can synthesize images so that they have smooth gradation changes that are similar to the nature of the blurring of pixels near the contours of natural images. There wasn't. The present invention was made in order to solve the above-mentioned conventional problems, and it is possible to preserve the state of color change near the contour line in the original image for the combined image. It is an object of the present invention to provide a method for creating a cutout mask that can obtain a close contour line quality and, by extension, a well-blended, natural-looking composite image.

[Means to solve the problem]

The present invention provides a method for creating a cutout mask for creating a composite image by performing arithmetic processing on original image data and using electronic cutout and composition processing. Then, create an evaluation function that indicates the possibility that an arbitrary color belongs to the real part, specify the area that spans the background part and the real part from which you want to cut out the image, and then, for all pixels within the specified area, Substitute the pixel value into each of the evaluation functions, determine the synthesis ratio to be used when compositing the background part and the real part based on the value of each evaluation function obtained by the substitution, and calculate the calculated synthesis ratio. The above-mentioned problem is solved by creating mask data having gradations made up of values.

[Effect 1] In the present invention, in the cutout mask creation method, evaluation 1iillil indicating the possibility that an arbitrary color belongs to the background part
For the evaluation function that indicates the possibility that an arbitrary color belongs to the real part, substitute the pixel value (for example, pixel density) in the area where you want to cut out the image, and calculate the value of each evaluation function obtained by the substitution. The composition ratio is determined based on the value. According to the evaluation function, for pixels that are located near the outline of the original image and have an intermediate color between the background part and the substance part, the ratio of the background color and the substance color is mixed to determine the color of the pixel. You can roughly guess. Therefore, the synthesis ratio reflects natural density changes near the contour line. Therefore, when cutting out the real part and placing it on another background image and compositing it, the original image (image to be cropped) is set at a compositing ratio that follows the natural floor changes of the original image at each pixel near the contour line. Since it is possible to mix the background image with the background image, it is possible to preserve almost the degree of color change near the outline in the original image in the combined image. Therefore, it is possible to create a blended, natural-looking composite image that has contour lines with smooth gradation changes that are similar to the nature of blurring of pixels near the contour lines of a natural image. Further, step-like wobbling does not occur in the contour line in the diagonal direction with respect to the scanning direction. Embodiments Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. The configuration of a cutout mask production system according to this embodiment is shown in FIG. As shown in FIG. 1, this system mainly consists of an image data input section 10, an image data storage memory 12,
, position input unit 14, training area storage memory 16
, cutting area storage memory 18, evaluation function creation section 20, evaluation function storage memory 22, mask density calculation means 2
4, a mask data storage memory 26, and a mask data output section 28. The image data input unit 10 is for inputting pixel values of images to be cut out and combined, such as pixel density data (a set of this data is referred to as image data). Cyan (
C), magenta (M), yellow (Y), I (Bk
) to input image data consisting of each color density value.
A scanner can be used. The image data storage memory 12 is for storing the input image data. The position input unit 14 is configured to perform a position input process on an image for sampling image data in order to create an evaluation function indicating the possibility that an arbitrary color belongs to the background part and an evaluation function indicating the possibility that an arbitrary color belongs to the substance part. area, i.e. training area;
This is for indicating the target area, that is, each area of the cutting area, for which cutting is desired. This position input section 1
For example, a pen recorder, a mouse, or a digitizer can be used for 4. The training area storage memory 16 and the cutting area storage memory 78 are for storing the positions of the input training area and cutting area, respectively. The evaluation function creation unit 2o creates evaluation functions, for example, Mahalanobis distances to be described later, as many as the number of training Airies based on each pixel value of the input training Airy. This evaluation function is the evaluation 1ifi mentioned above.
It is stored in the function storage memory 22. The mask density calculation unit 24 uses the evaluation function to
This is to create gradation mask data for each pixel in the cutting area. The created gradation mask data is stored in the mask data storage memory 26, and the stored mask data is outputted from the mask data output section 28 as necessary for cropping the image data. The effects of the embodiment will be explained below. The cutout mask creation system according to this embodiment creates gradation mask data according to the procedure shown in FIG. That is, first, image data is input from the image data input section 10 and the image data is stored in the storage memory 12,
If necessary, it may be displayed on a display means (not shown) (step 101). Next, the operator selects whether the position to be input from the position input unit 14 is in the training area or in the cutting area (step 102). However, at the beginning of this procedure, it is necessary to create the evaluation function f (γ〉) in advance, so it is only possible to proceed to step 102 and input the training area. After entering the information, you can select the cutting area in sequence.Furthermore,
The training area can be changed as the cutting process progresses. Then enter the training area (step 10)
3). Please note that when entering this training area,
This can be done in a variety of ways, one of which is
Input one or more training areas each from the background part and the real part. Next, an evaluation function is created for each training area based on the input pixel values of each training area. As the evaluation function, for example, the Mahalanobis distance shown in the following equation (1) can be used. f (y)2=(x-y) −Cov-” (y-
y)...(1) However, x is the mode average (vector) of the training area, and COv is the covariance matrix of the concentration of the training area. Further, t represents a transposed matrix of <n-v). This Mahalanobis evaluation function f(y) always takes a non-negative value, and the smaller the value of the evaluation function f(V), the closer it is to the color distribution of the training area. Note that another method in steps 103 and 104 is to take one or more training areas spanning the background part and the real part, and perform clustering by regarding the obtained training areas as one area. Then,
A possible method is to determine whether each divided cluster belongs to the background part or the substance part, and use the Mahalanobis distance from each cluster determined to belong to the background part and the substance part as the evaluation function. Either way,
As a result, one or more evaluation functions are created for each of the background part and the substance part. Next, enter the cutting area (step 1)
05),. Next, using the evaluation function, the mask density (represented by gradation) is calculated for each pixel in the cutting area, and mask data having gradation (N-tone mask data) is created (step 106). This gradation mask data is created, for example, as follows. For each pixel in the cutting area, the color of that pixel (
Pixel value) is assigned to each evaluation function, and the one with the smallest value among the evaluation function values belonging to the background part is determined as the distance from the background part
Similarly, for the real part, the minimum value among the evaluation function values is set as the distance Y from the real part. Using these distances X1Y, the mask density parameter θ of the pixel is calculated using the following formula (
Calculated according to 2). θ−arctan(Y/X) −・・−”・＜
2) On the other hand, the minimum value θmin of the parameter that changes the mask concentration between 0% mask and 100% mask,
The maximum value θmax is set in advance so as to satisfy the relationship expressed by the following equation (3). Okuθmin <θalax <yr/2 − (3
) The mask density is determined from the following parameters (I) to (I[l) from the set values θ1n1θWaX of the parameters θ, and the mask density is used as mask data. (I> If θ<θ1n, set the mask density to 0%. <II) If θ>θmax, set the mask density to 100%.
%. (I[[) When other than the above (I>, (II) (θwax
>θ>θsin), the mask density is determined from the relationship of the following equation (4). ((θ−θwin) / (θmax−θ1n))X
(100% mask) ・・・・・・・・・ (4) The relationship between these (I) to (I[[) is as shown in Figure 3.
It is represented by a change in the parameter θ in a plane consisting of distances X and Y. Also, as shown in Figure 3, with a 0% mask, only the color of the real part is used as the pixel color, and as the density increases, the ratio of the color of the background part increases, and with a 100% mask, only the color of the background part is used. The color of the pixel. Next, the gradation mask data is displayed on a display means (not shown) (step 107). If the displayed mask is inappropriate, the process returns to the light step 102 to continue the process again, and if it is appropriate, it is stored as gradation mask data (step 108). Appropriate gradation mask data can be created as described above,
By combining the real part image and the background part image using this gradation mask data, it is possible to obtain a composite image in which the contour line blends in well like a natural image. Note that after determining the density of the mask for all pixels in the cutting area as in step 106, the following processing can be performed as necessary. That is, all pixels in the cutting area are sequentially targeted, and if there is no 100% mask within a certain range from each target pixel, the mask density of the target pixel is set to 0.
You can perform processing to rewrite it to %. This is a process that limits the blur width (width of a region where pixels with halftones other than 0% and 100% exist) to within a certain value. At the same time, it is also a process of removing an isolated 100% mask by regarding it as an error. In other words, depending on the image, there may be a color close to the background part inside the real part, and a halftone mask is applied unnecessarily to the part where such color exists, or a 100% mask is applied to scattered dots. This may cause some inconvenience. In such a case, the above-described process is effective to some extent, and unnecessary masks can be removed. Note that the above processing can be effectively performed not only immediately after the mask density is determined in step 106, but also when it is performed on all pixels immediately before step 108. In the above embodiment, the Mahalanobis distance of formula (1) was used as the evaluation function, but the evaluation function of the present invention is not limited to this, and may be applied to other methods, such as Japanese Patent Application No. 1-340452 already proposed by the applicant. Any evaluation function can be used, such as the Euclidean distance shown in 2003 or 2-6647. [Effects of the Invention] As explained above, according to the present invention, it is possible to obtain a cutout mask that can be synthesized by preserving almost the state of color change near the contour line in the original image in the synthesized image.
The synthesized image has contour line quality close to that of the original image. As a result, the composite image blends in well and looks natural, and the excellent effect of eliminating the step-like jitter that occurs in the contour line diagonal to the scanning direction can be obtained.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing a conceptual configuration of a cutout mask creation system according to an embodiment of the present invention, and FIG. 2 is a flowchart showing a mask data creation procedure for explaining the operation of the embodiment. FIG. 3 is a diagram for explaining a method of determining mask density based on the distance from the background part and the real part. 10... Image data input section, 12... Image data storage section, 14... Position input section, 16... Training area storage memory, 18...
Cutting area storage memory, 20...Evaluation function creation unit, 22...Evaluation function storage memory, 24...Mask density calculation unit, 26...Mask data storage memory, 28...Mask data output Department.

Claims (1)

[Claims] (1) In a method for creating a cutout mask for creating a composite image by performing arithmetic processing on original image data and using electronic cutout and composition processing, an evaluation function indicating the possibility that an arbitrary color belongs to the background portion; Create an evaluation function that indicates the possibility that an arbitrary color belongs to the real part, specify a region that spans the background and the real part from which you want to cut out the image, and calculate each of the above for all pixels within the specified region. Substitute the pixel value into the evaluation function, determine the synthesis ratio to be used when compositing the background part and the real part based on the value of each evaluation function obtained by the substitution, and from the calculated synthesis ratio value. A cutout mask creation method characterized by creating mask data having gradation.